Dennis A. Clements

Economic Analysis of Influenza Vaccination and Antiviral Treatment for Healthy Working Adults

Context Strategies to decrease the adverse consequences of influenza include vaccination and antiviral therapy. No previous study has compared these two strategies in healthy working adults. Contribution In this costbenefit analysis, vaccination strategies resulted in higher net benefits than strategies that did not include vaccination. The health benefits of most antiviral treatments equaled or surpassed their costs. Clinical Implications Vaccinating healthy working adults against influenza is an economically attractive strategy for preventing the adverse consequences of influenza. Antiviral treatment for persons infected with influenza also saves money, but head-to-head comparisons of the available therapies are needed to define the most cost-effective regimen. The Editors Each year, influenza affects 10% to 20% of the U.S. population (1). In high-risk populations, such as elderly persons, influenza causes up to 20 000 deaths per year (2). Even in young healthy persons, influenza significantly affects direct health care costs, losses in worker productivity, and quality of life (3). In terms of therapy, yearly vaccination can reduce the risk for influenza, and various antiviral medications (for example, amantadine, rimantadine, zanamivir, and oseltamivir) can decrease the duration of illness for a person with influenza. However, yearly vaccination of healthy adults is not absolutely recommended, and it remains unclear whether the benefits of anti-influenza medications justify the costs (4, 5). We compared the costs and benefits of contemporary preventive and treatment strategies for influenza in a sample of healthy working adults. We conducted our study as a costbenefit analysis because most effects of influenza in a healthy adult sample (work-days lost and symptoms) are noncatastrophic. Using a decision model, we compared competing strategies by incorporating influenza vaccination versus nonvaccination and antiviral therapy (zanamivir, oseltamivir, rimantadine, and amantadine) for infected patients. This decision model considered the direct costs (for example, medication costs) and indirect costs (for example, lost wages) associated with each treatment strategy. In addition, our model incorporated the patient-determined relative value for relief from influenza symptoms and for avoiding medication side effects. To measure these variables, we used survey data and a conjoint analysis by using a utility valuation approach. Finally, we used sensitivity analysis to identify factors that could affect the optimal strategy. Our work adds to previous studies in considering antiviral strategies for influenza infection. Methods Model Overview We used a decision tree to model the choices of whether to vaccinate and whether to treat influenza (if influenza infection occurred) with one of several agents (Figure). The decision model was constructed for healthy persons 18 to 50 years of age without any significant comorbid conditions. All costs and benefits were framed from a societal perspective, which we defined as the perspective on outcomes of an intervention that accounts for all health effects (harms and benefits) and all costs (regardless of whether a monetary transaction occurs and who pays). We used the following equation to calculate our costbenefit analysis: Figure. Decision tree showing the strategies for influenza prevention and treatment net benefit (cost) = benefits of vaccination and treatment costs of vaccination and treatment. The Appendix provides the details of the equation and model. In brief, we included in the model eight treatment optionsthe eight possible combinations of influenza vaccination before infection (yes or no) and antiviral therapy if infection developed (using rimantadine, oseltamivir, or zanamivir or no treatment). We initially considered amantadine therapy as a possible treatment option, but because amantadine has a higher incidence of side effects than rimantadine and amantadine was less efficacious, amantadine was dominated by rimantadine in all subsequent analyses and was excluded. Because we were considering a healthy young population, we did not assume that vaccination or antiviral therapy would affect mortality or provide any long-term health benefits. Although these assumptions are conservative, they are consistent with the results of previous trials in healthy young persons (3, 6). We programmed the model using DATA software, version 3.5 (Treeage Software, Williamstown, Massachusetts). Probabilities Table 1 shows the value estimates used in the base case and the ranges evaluated in our sensitivity analysis. For a healthy adult, the probability of contracting influenza during an influenza season has been estimated to be 15% (range, 1% to 35%) (6-9). Vaccine efficacy for preventing influenza infection was estimated to be 68% (range, 50% to 86%) (6). Because rimantadine is effective only against influenza A and because oseltamivir and zanamivir are each effective against influenza A and influenza B, the prevalence of influenza B is important in determining the optimal antiviral therapy. The baseline prevalence of influenza B among influenza strains in a given year was assumed to be 16.3% (range, 1% to 86%) on the basis of the average yearly rate for prevalence of influenza in the United States over the past 10 years (Unpublished data). In terms of side effects, we considered the probability that rimantadine caused side effects of the central nervous system (characterized as dizziness, nervousness, and anxiety) in 2% of patients and gastrointestinal side effects (characterized as nausea) in 1% of patients (2, 3). For oseltamivir therapy, the probability of gastrointestinal side effects (characterized as nausea) was considered to be 9% (29). Because significant side effects rarely occur with use of influenza vaccine or zanamivir, the probabilities for these variables were not included in the base-case model (3, 6, 9, 13). Finally, on the basis of a previous study (14), we assumed that 17% of patients who developed influenza infection would receive antibiotic therapy (at a drug cost of

Postlicensure study of varicella vaccine effectiveness in a day-care setting.

17.50) (12). This figure was reduced to 11% in infected patients who received antiviral medication, based on data from a previous study (14). Table 1. Base-Case Values and Ranges Costs All costs and benefits are expressed in 2001 U.S. dollars. We calculated these dollar figures on the basis of 2001 medical cost and wage index data (28). The cost of vaccination, including the cost of administration, was

Over five-year follow-up of Oka/Merck varicella vaccine recipients in 465 infants and adolescents.

10.41 (15); a 5-day course of rimantadine therapy was

Haemophilus influenzae type b infections in Victoria, Australia, 1985 to 1987

17.50 (19), and 5-day courses of zanamivir and oseltamivir therapies were

Comparison of the epidemiology and cost of Haemophilus influenzae type b disease in five western countries.

47.50 and

Virulence Associated with Outbreak-Related Strains of Burkholderia cepacia Complex among a Cohort of Patients with Bacteremia

57.22, respectively (30, 31). The cost of a physician visit was assumed to be

Interference of Antibody Production to Hepatitis B Surface Antigen in a Combination Hepatitis A/Hepatitis B Vaccine

17.00 (Current Procedural Terminology code 99201, office or other outpatient visit) (range,

Concurrent administration of inactivated hepatitis A vaccine with immune globulin in healthy adults

15 to

MODIFIED VARICELLA-LIKE SYNDROME

40) (17, 18). Patients were assumed to incur these costs when seeking treatment with any of the antiviral therapies. Two recent trials reached different conclusions on whether vaccination saves the cost of physician visits. Because the vaccines come from different years, the vaccines in these two trials had nonmatching antigens and differed in terms of efficacy (6, 9). To be conservative about the benefits of vaccination, we assumed that vaccination did not save costs from physician visits. For antiviral therapy, we initially assumed that medication-related side effects would not require additional physician office visits. We tested this assumption by modeling a physician visit for each medication side effect experienced and found that results on optimal strategy were unchanged. No trial has demonstrated that vaccination or antiviral therapy decreases hospitalization rates for influenza-related illness (3, 6) Benefits For each episode of influenza infection, an average of 2.8 workdays are lost (21, 22). Wages gained by avoiding influenza because of vaccination therefore equaled 2.8 days times an average daily wage of

Varicella vaccination in children.

142.10 (16), or